Team: Michael Gobreski, Joseph Karbowski, and Shivangee Makharia
Receiving and attending to consumer input, especially complaints, is crucial to any business. Our capstone project with CGI and PNC aims to create a program that generates support tickets based upon complaints received from consumers. The support ticket will be comprised of two components: the department who can address the issue and some additional information about the issue so that the department can address it. We use contemporary machine learning techniques to achieve these goals. We train a state-of-the-art BERT model on the Azure platform to classify and analyze the sentiment of consumer input, and we construct a web scraper to extract consumer input from Twitter and other social media platforms to streamline the creation of support tickets from consumer input. Ultimately, this capstone gives us rich and valuable experience working with big data and using natural language processing, deep learning, and transfer learning to solve a real problem that companies face.